If the word “survey” makes you want to shudder, you’ve come to the right place. Building a survey that doesn’t suck for your UX research project is pretty tough. First of all, surveys are very tempting to use because they get you a lot of numbers and they verify your assumptions about your product. However, those are also the downfalls of surveys – they verify possibly false assumptions and can get you “good” numbers that back up those made-up assumptions. And that sucks. This post breaks it down and helps you put together a survey that doesn’t suck.
Why am I writing this? A few months ago, I was tasked with running a survey to collect data about message editor usage for AWeber. I had no idea how to properly conduct a survey and so a search began on the best way to tackle this beast. This post is a short collection of tips on how to put together a survey that doesn’t suck.
Before we get started
Surveys are a great supplementary method to other qualitative tests. Surveys can collect both qualitative and quantitative data.
Writing your questionnaire
A survey != questionnaire. A survey is the method and a questionnaire is the actual thing used to ask the questions. Got it? Good! This post will outline some best practices on writing that effective questionnaire.
Set a Goal
This should come as a no-brainer, but figure out what you want to find out first! Determine, based on this goal, whether a survey is the most appropriate method of research. A few months back, our UX team at AWeber was given the task of finding out the biggest reasons why people were using a previous version of our message editor. A survey was appropriate because: 1. we were looking for quantitative data and 2. we already had qualitative research from which we drew our theories and assumptions.
Collect your research
Gather your relevant user interviews, usability studies and whatever other qualitative research you may have. Comb it for possible answers to your initial questions. As Caroline Jarrett states a relevant, and therefore a well researched question, creates trust between the researcher and the research participant.
I like to write all possible answers onto sticky notes, and then, card sorting them into relevant groups. I prefer writing my answers first and going all Jeopardy on them by writing my questions after I have gathered relevant answers. At this point after you have your groups of answers, you have, what seems like a mess of information. You have answers to questions not yet written in seemingly relevant groups. So, what kinds of questions should be asked?